I've been using MantisBT since 2005 — and I have to admit I only discovered this forum recently. Better late than never.
I wanted to share something that genuinely changed how our team interacts with Mantis daily. Over the past year, AI coding assistants have become a core part of our workflow, and we got frustrated constantly switching context: read a ticket, copy details into the chat, fix the code, then go back to Mantis to add a note.
There were already a couple of MCP servers for MantisBT out there, but none of them quite fit our needs — and most hadn't seen active development in a while. So I ended up building my own.
MCP (Model Context Protocol) lets AI tools like Claude Code talk directly to external services — meaning the AI can read, search, create, and update your Mantis tickets natively, without you ever leaving your editor.
One thing that pushed me further was a known REST API limitation discussed here not long ago: there's no native way to do keyword searches programmatically. The only workaround is fetching hundreds of issues and filtering client-side — not the approach we wanted to rely on. Our MCP server includes an optional local semantic search index that addresses exactly this — working entirely offline, without touching the API, and supporting both keyword and natural language queries.
It sounds like a small thing, but it entirely changed how we handle bug-fix sessions. The AI has full ticket context, can create follow-up issues, add notes when done — all while you stay in flow.
Code: Select all
npx @dpesch/mantisbt-mcp-serverCurious if anyone else here has been exploring AI tooling around MantisBT — or if you have questions, happy to answer.